Special Seminar: System Identification for Spaceflight Applications
Frances Zhu
2:30 PM on 2019-02-21
3-180 Keller Hall
Abstract:
Flux-pinned interfaces for spacecraft are an action-at-a-distance technology that can maintain a passively stable equilibrium between two spacecraft in close-proximity using the physics of magnetic flux pinning. Although flux pinning dynamics have been studied from a material-science perspective and at an interface level, there is a need to better understand the sensitivities and implications of system-level designs on the flux-pinned interface dynamics, especially in designs with multiple magnets and superconductors. These interfaces have highly nonlinear, coupled dynamics that are influenced by a multitude of physical parameters. A general flux-pinned dynamic model framework is outlined to predict dynamics between two spacecraft governed by flux pinning physics. Spaceflight and experiment architectures are developed for active, contactless spacecraft interaction. From these architectures, a series of experiments are conducted in different environments, ranging from extremely flat floors to microgravity flight. Dynamic data collected from these experiments inform the interface’s system capabilities and progress predictive dynamic model refinement of the physical system. In the form of reinforcement learning using genetic algorithms, robots can identify their own system dynamics and an optimal control policy in real-time, and in complex, extreme environments. Ongoing work includes continued system identification with neural network approximation, which includes theory in neural net interpretation.
Bio:
Frances Zhu earned her B.S. in Mechanical and Aerospace Engineering from Cornell University, Ithaca in 2014 and is currently pursuing a Ph.D. in Aerospace Engineering at Cornell. Since 2014, she has been a Research Assistant with the Space Systems Design Studio, specializing in dynamics, systems, and controls engineering. Her research interests include flux-pinned interface applications, spacecraft system architectures, robot dynamics, estimation, controls, and machine learning. Ms. Zhu is a NASA Space Technology Research Fellow.